sup8pdct

Just for giggles and something to do while I am waiting for my broken right arm to heal, I decided to have a close look at DosXE because the docs says it can work the 16meg drives but there was nothing to set one up. I have always woundered about that since it was released. The docs don't go into any detail about drive type and how to make different ones. So with all my now spare time doing nothing for next 6 or so weeks I dissassembled dosXE and had a look. Discovered several things and worked out the drive type table. DosXE can work with 16MEG drives BUT I had to set it up manually with disk editors etc and used it on my SIO2USB. Drive type table and VTOC info is below for who ever has any interest. Will be amazed if anyone askes for a blank 16meg image that I have made. DosXE drive type table. 8 in total. In memory, starting at $D800. Also written to disk. 5 setup as standard are AT810 :SSSD, AT1050 :SSED, XF551 :DSDD,130RAM and SSDD. 1-6 ;asci name eg:'XF551' unused bytes set to $00 7 ;number of sectors in Vtoc 8 ;? could be number of sectors per bit of Vtoc. always set to $01 9-10 ;number of sectors on disk +1. max = $FFFF. LO-HI 11-12 ;number of usable (free) sectors after format. LO-HI ;Single/enhanced density has the real numbers /2. ;DosXE uses every block as 256 bytes so single and enhanced ; are treated as 2 sectors per block 13 ;First byte of VTOC map 14 ;first directory sector. always right after last vtoc sector =byte 7 +4 15-16 ;SIO routine to use for data block. LO-HI 17-18 ;Sio routine to use for command frame LO-HI 19 ;Read cmd. usally $52. xf551 has high bit set 20 ;Write cmd. can be $57 or $50. xf551 has high bit set 21 ;Speed byte for command frame 22 ;speed byte for data frame 23 ;format command. xf551 has high bit set 24 ;format time out value 25 ;percom block: number of tracks 26 ;percom block: step rate 27-28 ;percom block: sect per track hi-lo 29 ;percom block: number of sides -1 30 ;percom block: 0=FM 4=MFM 31-32 ;percom block: bytes per sect HI-LO Sio routine entry points used by dos XE. $D5D uses rom sio routines after enabling os rom. Use for PBI devices speed is always 19200 unless PBI devices or custom OS with high speed sio is used. Ignores speed byte in drive type table. $FE5 High speed sio routine. Uses Speed byte. Doesn't jump to PBI devices $DF32 Entry point for data frame for 128 byte sector drives only. always uses rom sio $13F7 Entry for ram disk data frame $1445 entry for ramdisk command frame XF551 is set to use DSDD only. uses $D5D for command, $FE5 for data. DosXE uses XF551 highspeed sio for DSDD only. XF551 can be set to at810 and AT1050 Dosxe cannot set up the vtoc properly above 8112 sectors. Dosxe locks up the computer if drive size is 65535 sectors during format. If drive type asci string doesn't match what dosxe has in it's table, drive will not be read If percom block doesn't match what what dosxe has in its drive type, drive will not be read. Exception for real unmodifyed 810 and 1050 drives. only asci string is used. This affects SIO2USB.SIO2USB percom block always has 1 for number of tracks regardless of drive size. VTOC 1st sector setup 1-6 ;copy of drivetype table 7-12 7-8 ;file number to use for next new file -1 LO-HI 9-10 ;Random number used to id drive. unique for each disk. both numbers must be different. 12 ;byte 13 of drive type table, 1st VTOC byte 13- end ;rest of VTOC. DosXE sets all to $FF for as many as needed Random number is used through out disk as part of sector labels SETUP.COM file doesn't display sector number properly for sector count above $8000. displays a negative number. James

Idont know about the windows version, but the dos version (unregistered) one cannot write at high speed. Have you tried forcing std speed for writes on the atari side? james

Check out this site. http://www.jsobola.republika.pl/schematy.htm Much usefull information find, you will James

I did a little snooping and found some answers for all possibly? C and J are A14 and A15 respectively (the text is wrong about which one). U9 is ANTIC and C/J are going to pins 19 and 20 respectively going under the chip to make those connections. The jumpers just bring those lines into the addressing scheme so that the full 64k is usable. And then on the 800XL, pin 8 of the 74ls51 is the CAS signal and since on the 600XL the CAS signal is derived from pin 8 of the 74ls32, can I suggest that the full 64k would not have an adequate CAS signal if it was derived from the 74ls32 instead of the 74ls51? THIS would explain why the PBI addon memory expansion is not fully compatible with the 800XL, the needed CAS signal is ONLY available inside the cover of the 600XL. This would have primarily to do with the MMU selecting RAM under ROM when set to do so by port b settings. There is an error in the diagram. U14 is actually a 74ls14 chip and NOT yet another 74ls158. The fact that U9 is ANTIC should not have been left out, just my 2 cents. Thanks for sharing the upgrade info, I thought I had them all before this especially those to be found at Craig Lisowski's site active so long ago. I wonder if this upgrade suffers from long memory troubles where it won't coldstart with a simple off/on cycling of the power switch due to the longer memory retention times of the much better silicone to be found in the 4464 chips? I suspect, yes it does exhibit extreme memory lag. I've seen some poor souls hang onto the power switch for a 10 count only to NOT get the desired coldstart but a warmstart right back to the frozen screen that they are praying to escape from someday soon. That look of pure exasperation can't be bought. Very close. The CAS signal comes from an or gate on the LS32. a 2nd gate on the same chip has A14 and A15 as it's inputs. the output is high if either or both A14 ,A15 are high ie any address 16K or higher. the first gate has CAS and Greater then 16K in and CAS select out. for address from $0 to $3FFF, cas is normal. Any other address, cas is disabled. The wire (x-y) bypasses this little circuit so the full 64K can be accessed internally. Cas on the PBI on the 600xl cannot come from the 74LS32. CAS needs to (and does) come from pin 8 of the 51 for the PBI. If there was a timing issue with the 1064, it would show up anywhere past 16K, Not just under the roms. The schematic I downloaded doesn't show where cas comes from for the PBI. MMU timeing difference for normal ram and ram under roms would be so very small I would doubt if it could be measured. James

Maybe yours friends one had some sort of problem. The built in ram check only checks the first 48K. Every one I have tried has worked 100% with 64K programs Other ram upgrades for the 600XL I have seen has the LS32 chip removed and a jumper placed in pins 8 and 10 of the chip socket which is the same as the wire from point x to y. James

All that is on the 1064 is a buffer for the datalines which is enabled when ram address is required and the address is above $4000, address multiplexers and delay buffers. oh and 64K of ram The MMU in the XL handles everything else. James

Very cool, nice finds, and thanks for the advice! And just to clarify, the XEGS has the most amount of memory of any Atari 8-bit platform, and doesn't have AV-or am I wrong on all that? The XEGS has 64K of ram, same as 800XL, 1200XL, 65XE and 800XE. the 130XE has 128K of ram out of the box. The XEGS however has the largest rom, 32K which has the os, basic and the game. XEGS has AV and audio out on seperate RCA jacks. James Ok, so what is the advantage to having that 32K of rom, and the os and basic in the XEGS? I would think that it would make it compatible with more games, because you wouldn't have to upgrade the memory-am I wrong? I all comes to board space and manufacture. Why use 3 chips(os, basic and game) with all it's assoated trackwork and switching when it can be done with one and a slightly modifyed MMU. Less holes to drill, less trackwork, smaller board etc. Otherwise, there is no advantage whatsoever. There are games out there that need more then 64K but there isn't all that many. James

Maybe a bad connection to the CPU board, a bit of crud somewhere where it shouldn't be or maybe the Ctia/Gtia needs to be reseated or replaced. James

Another cart to try is Star Raiders as it also looks like a diagnostic cart to the OS. Don't think the OS ROM is your problem. You can try replacing those two ICs but if you're getting the game to come up like that, I doubt they're the culprits. Trying plugging in one memory module at a time and power on. While not impossible, it's highly unlikely all three 16K modules are bad (I assume you have 48K). You could also have a bad PIA. I've got a couple of Star Raider carts, will try them later! Yes, she is 48K, will try a combo of the RAMS to see if that alters anything..fingers crossed P.S. Excuse my ignorance, what is the PIA? PIA is the 6520 chip looks after the joystick ports. James

Very cool, nice finds, and thanks for the advice! And just to clarify, the XEGS has the most amount of memory of any Atari 8-bit platform, and doesn't have AV-or am I wrong on all that? The XEGS has 64K of ram, same as 800XL, 1200XL, 65XE and 800XE. the 130XE has 128K of ram out of the box. The XEGS however has the largest rom, 32K which has the os, basic and the game. XEGS has AV and audio out on seperate RCA jacks. James

Is news to me. The 1064 plus 600XL = an 800XL right down to the extra 16K under the roms. I also had a friend with this combo and it worked with everything that worked on the 800XL tho it did flake out at times due to the contacts. James

How? At the moment it's really purple looking. Tempest The colour wheel on the 800 is easy to access. Remove the top cover and looking from the rear, you will see a hole in the alloy RF shield. Just jam something into the interlock switch and adjust away. James

Personally, I would try swaping the chip marked 6520 on the mother board. Not sure what the atari part number is.... James

I can say that card 2 is the cpu card. one that has a normal 6502 installed, not atari's special version. With this card, you can get a 65C02 and swap it with the orignal. Does this card have a crystal on it? James

Interesting... I didn't know that SuperDos was incompatible with SIO2PC. I can successfully boot SuperDos 5.0 from APE, but SD refuses to recognize any APE drive other than D1:. It would be interesting to know why it fails with other than D1: . Off the top of my head, I would suspect that APE responds too quickly, but that wouldn't explain why it recognizes D1: (?) But since SD doesn't support hard drives, that limits its usefulness for many users these days. Anyway, to your original question. If you want a MyDos ramdisk to *exactly* represent a 90K or 180K drive, then you would need the CSS Ultra Speed Plus OS. They may still have them for sale. I used that OS in a 320K XE for years. But I don't know about it supporting the 1200XL or XEGS. From memory: the default ramdisk for USP+ is D4: as a 720 sector SD disk, but it can be changed to D9: (or other) and/or changed to DD provided you have the ram to support it. It is a proprietary OS module and must be installed with several solder points. Not terribly difficult to install, but not "plug and play" either. Here's the link, if you find the USP+ of interest: http://www.nleaudio.com/css/ -Larry There are other bugs in SD if you have a XF551 with a HyperX mod you can't use SD and SD can't read a 360k formatted disk by Mydos even though they have the same numbers of sectors MyDos can't read a SD 360K formatted disk! I wouldn't be mad if someone made mods to SD. Double sided in superdos is setup similar to enhanced density to make things compatiable with all the dos 2 clones that were around in 1986. Also superdos was set up to use an unmodifyed xf551 and was written to look for it and it's density change bug. At the time, there was no harddrives ,atr8000 etc in australia so no support was written in to it. It would take a major rewrite to support subdirectories, a minor one to support harddrives etc. I do have the source code which is written in synassembler and next to no comments. Paul Nicholls was a talanted programmer and he had a working Zmodem protocal till he hit the turps and formatted his floppies. He also wrote a 8K monintor program for the 800 that resided in 2 banks at $C000 - $CFFF and other projects One trick in superdos is if files have the extention RAM on the boot disk, those files are copied to ramdisk on boot up. James

Well this is a WIP, and have only been able to spend a little bit of time on it-- so it progresses slowly when it does like most hobby projects. I re-did the design and at this point I have essentially an all-in-one board that can switch back and forth between different three different 800 memory maps (stock 48K, stock 52K, Axlon 512K) and four XE memory maps (stock 128K, 320K, two different 576K). I've had to physically re-work the prototype to allow a larger CPLD to allow for this extra functionality. The point is, it can be done. While the hardware is essentially the same, there are some aspects with the design of the 800 that warrant some extra attention. The simplest option to turn it into a 64K 800XL is to equip your OS board (I've done this in the past and was part of the initial proof-of-concept) with an additional 16K of memory and put the XL OS in ROM with a connection to PIA bit 0. That would provide the minimum compatibility, but you'd still have work to do on the reset circuit, cartridge interlock, and fix a pin on the cart slot (the 800 doesn't supply the phase 2 clock to the cart slot, only RAS). I don't know what level of experience you have soldering or what equipment you have, but it can get rather involved. Sounds goooooooooood!!!!!! Can't wait.............. James

Your track zero sensor is broken. on the mech, the round object on the top right is the motor. the object directly below is the sensor. It has 4 wires going to it, it is sitting loose off to the left of where it should be. Just maybe, you can glue it back to where it should go. The small protrusion on the head slide should go in the gap of the broken part. Yell out if you want pics of what one should like. James

I have a least one Pal machine that has those 2 wires. It all depends on where it was made and what board revision. As for differences, Look for an extra crystal, an extra chip (74LS74) and some support components. All Pal XL's I have seen don't have the channel select switch despite support for one on the board and the modulator. Other differences are the 2 graphic chips and the main crystal. James

I would still clean the head even tho it looks clean. I have had instances where it looked clean but had a bit stuck on it. I would ask nicely someone who is near to you on this forum to send you a disk or 2 with dos on it to make sure your drive is ok. With your current disks, some times they get a coating that dirtys the head that once gone, will read ok. Could take several reads and cleans before they come good. James.

This looks like a 48K board to me, as it contains 24 pcs 16k*1 DRAM (presumably 3 banks making up for 16K each). Typing in ? FRE(0) in BASIC will tell you the exact amount and if the memory is working OK, so my advice would be to obtain a BASIC cart. re-atari Can you please take a picture of the bottom of the mother board of the 400 and post it? James

SuperDos will work with it and I think SDX will also work. James

32K is max amount of ram that can go in with out soldering. A simple plug in just like 800 but with more work. . James,

32K is max amount of ram that can go in with out soldering. A simple plug in just like 800 but with more work. . James,

That is how my 800 keyboard is. Tho cannot remember how it worked. When I come back from my holidays (6 weeks) will take mine apart and take some pictures, unless someone beats me to it. James

How it the keyboard made? If it a type of mylar, clean both the mylar and the rubber under the key, Use contact cleaner. If it is switches, a squirt of contact cleaner may work. James It has freed up some but there are so many more to go. I think I am going to buy a new one. I found one at B&C for $20.00 new so I think I will go this route. I took the keyboard apart and really didn't see a way to clean it. Thanks, Mark